Among the host of features present in the first release of Asana Intelligence is the ability to auto-generate actionable subtasks, each potentially with descriptions and assignees, based on a task’s existing description and comments.
If helpful, you can add all generated subtasks to your task in one click, or you can first individually discard those you don’t need.
The best way to determine how this feature can help you is to try it in different circumstances to assess where it provides the most value. Let’s review one such attempt below to gain familiarity with the feature and see how fast and easy it is to experiment with.
My use case was “Start designing a new requests processing workflow,” so I entered a new task with that name and then briefly used voice entry to brainstorm some initial approaches in real time to get my thoughts into the task description:
Satisfied with the core approach, I used the Task actions menu to have Asana Intelligence try to derive meaningful subtasks:
It did a pretty darn good job!
By just typing A the subtasks were added:
It did an ideal job of translating my thoughts into actionable work, maintaining the order of effort, by following key best practices for (sub)task titles:
- Begin with an action word (verb) followed by an object
- Keep it short (but not too short)
- Make sure it’s easily scannable
It wisely tossed out a lot of connector words that are useful in prose (the description) but not so in task titles and changed the verb form as appropriate.
It cleverly excised my unnecessary addition after MVP prototype: “[. . .] to use to gather information.”
The only real bone to pick could be that it combined the first two interviews into a single subtask, although that may well be perfectly appropriate (and certainly easily editable if not).
This is a simple example, granted, but it has encouraged me to keep using this feature. I find it tedious to generate well-formed subtask titles, but comparatively easy to speak or write more off-the-cuff about general intent. I’ll be happy to start offloading the drudgery!
Trial and error is the only way to assess many AI features. But remember not to close the door too soon; I’m sure these features will evolve rapidly.